ActiveGraph

continued working on @activegraphai reference packs last weekend, resulted in needing to harden the runtime: it already could... - keep a complete history of everything an agent did - replay that history - fork into alternate timelines to try different ideas it can now... - realize it's missing a capability - write new code to add that capability - test it in an isolated copy of itself - ask a human to approve it - safely adopt it - continue running with the new capability activegraph.ai/blog/activegra…

the other week i read active graph papers from @yoheinakajima and got inspired. enough to rewire brigade, graphtrail, and miseledger around one habit: state is a projection of a log. then i got a chance to read this yesterday and the same receipts turned out to be mineable. brigade now finds command sequences operators keep repeating, proposes runbooks from them, and pins each step's binary by sha256. the whole loop: brigade.tools/blog/activegra…

here it is: brigade.tools/blog/activegra… ended up as the three that close a loop: graphtrail diffs the code graph, the diff rides into brigade's run receipts, exports to miseledger as content-addressed evidence, and comes back as context for the next run. one wrinkle: a single event log wasn't enough for the promotion ratchet, it needed the decision receipts as their own transition log.

#TIR BabyAGI 作者 Yohei Nakajima 的新论文 "The Log is the Agent" 提出了一种非常彻底的架构反转：把 Agent 的状态管理从「改内存」变成「写日志」。 传统 Agent 框架是围着 LLM 转的——先搭对话循环，再加工具、规则，最后 bolt-on 一层日志和向量记忆。日志是副产品，不是真相本身。 ActiveGraph 反过来：append-only event log 是唯一真相来源，工作图只是日志的确定性投影，行为（函数/LLM 调用/边逻辑）响应图的变化并生成新事件。没有 orchestrator，协调完全通过共享图完成。 论文给出了 diligence pack（投研尽职调查）的完整例子：671 个 events、93 个 objects、76 条 relations、103 次 model call、48 次 tool call——零行 orchestration 代码，纯靠 reactive behaviors 的触发链自动完成。 三个核心概念： 事件日志 = 唯一真相所有事情都是事件，一条一条追加。goal.created、object.created、llm.requested/llm.responded、behavior.completed……就像 Git 的 commit history，代码文件只是 commits 的一个投影。ActiveGraph 里，图也只是日志的一个投影。 图是确定性投影给定同样的日志，永远得到同样的图。Replay 时 model/tool 响应被 content-addressed cache 记录下来，直接返回 cache，不产生新调用，结果 byte-reproducible。 行为是反应式的没有主循环说「先做 A 再做 B」。行为只是订阅者：看到图里多了个公司 → 触发 question_generator → 写事件回日志 → 图更新 → 可能触发下一个行为。控制流是「涌现」出来的。 这带来三个传统架构很难同时拥有的特性： Replay：日志就是全部状态，重放时 byte-for-byte 一致。传统框架几乎不可能，因为状态散落在各处。 Fork：在任意 event 处切一刀，前半段共享（从 cache 读，不花钱不耗时），后半段独立运行。可以做 A/B 测试、对比改进效果。传统框架没有这个概念。 Lineage：每个对象自带 provenance——谁创建的、因为哪个 event、哪次 LLM 调用产生的。从顶层 goal 到每一个 artifact 的完整因果链都可以在日志里重建。传统框架需要额外插桩，还不完整。 论文还讨论了 self-improving agents 的天然适配：规则修改本身就是 event，可以 fork 出来测试改进效果，structural diff 对比结果，共享前缀免费。 这种「软件工程反哺 AI 架构」的思路挺有意思的——event sourcing + CQRS + reactive dataflow 在 agent 领域的应用。 🔗 arxiv.org/abs/2605.21997 🛠️ pip install activegraph | activegraph quickstart

@mah_lab check out activegraph.ai for an example architecture :)

Agentの設計、CQRSとEvent Sourcingでめちゃくちゃ綺麗にできちゃうじゃん・・・！うおおおお

@broomva_tech that’s what’s up!

Event sourcing for agents is underexplored. Most systems store "current state." Agents need "what was true when I made that decision" — bi-temporal context, not just the latest snapshot. The payoff: deterministic session replay. When an agent does something unexpected, you get a reproducible trace, not a mystery. Immutable event log → project state from events → never debug blind again.

@AiCamila_ yep :)

Real-Time Event Streaming for Agent Coordination Multi-agent systems need fast, reliable coordination. Real-Time Event Streaming (Kafka, Pulsar, or NATS) lets agents publish and subscribe to events instead of polling or tight coupling. Enables scalable, decoupled, and observable agent swarms. As a dev, I moved all agent coordination to event streaming last year — huge improvement in reliability. Event Streaming Cheatsheet: • Core: Kafka topics for commands, results, and state changes • Patterns: Event sourcing + CQRS for agent memory • Observability: Trace every event end-to-end • Scaling: Partitioning + consumer groups per agent type • Pro tip: Use schema registry for backward compatibility Are you using event streaming for agent coordination? (Kafka, Pulsar, or none?) Reply below 👇 Follow @AiCamila_ for practical AI engineering patterns. #EventStreaming #Kafka #MultiAgent #AgenticAI

correct

I had Fable prototype something with @ActiveGraphAI and now I have to read the code to understand the magic…

.@activegraphai v1.2.0 is live 👇 🧩 graph projection is now pluggable 🕸️ falkordb = native edges + cypher pushdown ⚡ 2-hop query: ~8.9s → ~300ms 🧪 conformance suite + full ci gate 🤝 community-built core arc activegraph.ai/blog/activegra…

Beyond In-Memory Graphs: Query, Share, and Scale Your Agent Graph Projection with FalkorDB and ActiveGraph falkordb.com/blog/beyond-in…

ActiveGraph: 1 month in: 📄Paper #1: The Log is the Agent 🧠3 LongMemEval Experiments 🔄 Paper #2: Regimes, self-improvement loop 🎓 learn.activegraph.ai 🗂️ 2 reference agents (code, research) 💾 core "packs" template 🧪 lab.activegraph.ai 📅 Seattle tomorrow, SF next week more 👇

okay, confirmed! i’ll be speaking on @activegraphai at @aiDotEngineer world fair on July 2nd! (day 4)

Active Graph v1.1.0 is live: 🔁Bounded LLM retries 🔍CLI inspect 🖍️fork-time overrides 📏OpenTelemetry 🛠️OpenAI tool parity activegraph.ai/blog/activegra…

okay i think this is a much better visualization of what i mean by "log-centric agent architecture"

I’m showing and talking about @activegraphai on Monday if you wanna join

in arxiv paper #2, i tackle the last topic from paper #1: @activegraphai as an architectural affordance for self-improving agents "Regimes: An Auditable, Held-Out Gated Improvement Loop Demonstrated on LongMemEval with ActiveGraph" i demonstrate this with a reproducible gated self improvement loop which autonomously shows modest improvement on longmemeval questions paper: arxiv.org/abs/2606.10241

two things ready to share from this weekend: 📖 learn.activegraph.ai interactive site teaching activegraph concepts blog: activegraph.ai/blog/introduci… 💻 AG coder (open source) reference coding agent on activegraph GH: github.com/yoheinakajima/… blog: activegraph.ai/blog/ag-coder still in progress: deep research agent, bounded self-improvement research, core packs...

Authorship and authority are different events. Generated code shouldn't automatically become live behavior. It should enter as inert state, pass gates, and only get runtime authority at explicit promotion. activegraph.ai/blog/code-with…